6. Workshop Alpine Hydrologie

Subsurface stormflow (SSF) is a runoff-producing mechanism that can substantially contribute to the stream’s storm hydrograph. Despite its importance, the hidden processes controlling SSF are still not well understood. To study SSF and characterize associated storage-discharge dynamics, we analyzed the recession behavior of multiple SSF events at three trenched hillslopes in a Black Forest catchment, Germany.  SSF triggered by natural and artificial rainfall was measured in three trenches (T1–T3). In addition to discharge (Q), water levels were monitored upslope of the trenches. We extracted SSF recession segments and evaluated a single linear reservoir (1LR) model, a two parallel linear reservoirs (2PLR) model, and a power-law relationship −dQ/dt = aQ^b. These models were selected because they are often used in conceptual hydrological models to simulate subsurface-flow contributions, including SSF.

Recession behavior varied across hillslopes: at T1, most recessions were adequately reproduced by the 2PLR model, whereas at T2 and T3 recessions generally followed 1LR dynamics. The 2PLR fits show that a transient apparent b > 1 can occur from the superposition of two linear reservoirs: b approaches 1 under clear fast- or slow-flow dominance, but steepens during the transition between the two reservoirs. Consistent with this mechanism, T1 had higher apparent nonlinearity (median b = 2.3) than T2–T3 (median b = 1–1.3). The comparison between natural and artificial rainfall events suggests that event-to-event variability in recession timescales is partly driven by changes in the upslope contributing area. At T1, water table dynamics indicate that the 2PLR behavior likely reflects drainage from two vertically distinct zones: an upper, high-conductivity zone represented by the fast reservoir and a deeper, lower-conductivity zone represented by the slow reservoir. At T3, variability in 1LR recession constant was partly linked to frequent double-peaked SSF events, in which the first peak drained faster than the second, suggesting that the second peak reflects drainage from a larger contributing area. At T2, most events were single-peaked and well represented by the 1LR model; less frequent double peaks showed a similar fast-to-slow drainage sequence. Overall, simple reservoir structures reproduced most recessions, but fitted parameters varied between and within events, suggesting event-specific changes in contributing area and/or drainage conditions.